<?xml version="1.0" encoding="utf-8" standalone="yes"?><rss version="2.0" xmlns:atom="http://www.w3.org/2005/Atom"><channel><title>Obsidian on GrevireBlog</title><link>https://blog.grevire.dev/en/tags/obsidian/</link><description>Recent content in Obsidian on GrevireBlog</description><generator>Hugo</generator><language>en</language><lastBuildDate>Fri, 15 May 2026 23:32:45 +0200</lastBuildDate><atom:link href="https://blog.grevire.dev/en/tags/obsidian/index.xml" rel="self" type="application/rss+xml"/><item><title>Behind the Scenes of GrevireBlog: How This Post Was Written by a Human+AI Workflow</title><link>https://blog.grevire.dev/en/posts/behind-the-scenes-grevireblog-ai-workflow/</link><pubDate>Fri, 15 May 2026 20:36:00 +0200</pubDate><guid>https://blog.grevire.dev/en/posts/behind-the-scenes-grevireblog-ai-workflow/</guid><description>Behind the Scenes of GrevireBlog: How This Post Was Written by a Human+AI Workflow Dictated by Brocasm and corrected and translated by mistral-14b
TL;DR — This article documents the human+AI workflow behind GrevireBlog: from an Obsidian inbox note (or a voice memo) to a published bilingual blog post. DeepSeek V4 Flash orchestrates the pipeline (drafting, Git, PRs), while Mistral 14B handles language correction and translation. Context grounding (Mem0, LLM Wiki, session memory) prevents hallucination, and human validation remains the final gate.</description></item></channel></rss>